Keynotes | K&Ts | GACs | Talks | Posters | Search

Poster C in Poster Session C: Wednesday, August 5, 9:30 – 11:15 am, Kimmel Center, Shorin & Rosenthal Rooms

Foraging minds: computational and neural foundations of perceptual flexibility

Claire Sturgill¹, Adel Movahedian², Abhilash Dwarakanath³, Vishal Kapoor⁴, Nikos K. Logothetis⁵, Theofanis Panagiotaropoulos⁶, Roxana Zeraati⁵, Shervin Safavi¹; ¹Technische Universität Dresden, ²Sharif University of Technology, ³German Primate Centre, ⁴Chinese Academy of Sciences (CAS), ⁵Max Planck Institute for Biological Cybernetics, ⁶National and Kapodistrian University of Athens

Presenter: Claire Sturgill

Adaptive behavior depends on flexible perception, the capacity to continuously revise internal representations in changing or uncertain environments. Perceptual multistability, in which perception spontaneously alternates between competing interpretations of an ambiguous stimulus, offers a unique opportunity to isolate the internal mechanisms that support such flexibility. Because the sensory input remains constant while subjective experience changes, these transitions reveal an active inferential process through which the brain updates its current hypothesis about the world. The broad presence of multistability across species further suggests that it reflects evolutionarily conserved computational principles. Yet its ecological and functional significance has remained unclear. Here, we propose that perceptual multistability emerges from an internal foraging strategy: an ancient normative mechanism that adaptively explores alternative interpretations to reduce uncertainty and mitigate risk. Using a reinforcement learning framework, we model each perceptual interpretation as an informational patch. The value of the currently dominant percept gradually declines over time, while the value of suppressed alternatives increases. A perceptual switch occurs when the expected reward of an alternative percept outweighs that of the current percept after accounting for switching cost, analogous to travel time in foraging. This simple principle captured a wide range of experimentally observed temporal signatures of multistable perception, including gamma-like dominance duration distributions, task-dependent modulation of perceptual dynamics, multi-timescale switching behavior, and reward-driven asymmetries in perceptual dominance. To identify the neural substrate of these value computations, we analyzed population spiking activity in the ventrolateral prefrontal cortex of monkeys engaged in a multistability task. Neural population dynamics in this region encoded the evolving values of competing perceptual patches in close agreement with model predictions. Together, these findings provide a unified mechanistic account of perceptual multistability as value-based internal foraging, linking ecologically grounded decision processes with higher-order cognitive functions and offering a principled explanation for flexible perception across species.

Topic Area: Decision-Making, Cognitive Control & Event Cognition